heliospheric observations of stereo-directed coronal mass ejections in 2008--2010

7/31/2019 Heliospheric Observations of STEREO-Directed Coronal Mass Ejections in 2008--2010

1/22

Solar PhysicsDOI: 10.1007/10.1007/s11207-012-0007-8

Heliospheric Observations of STEREO-DirectedCoronal Mass Ejections in 20082010: Lessons for

Future Observations of Earth-Directed CMEs

N. Lugaz1,2 P. Kintner2,3 C. Mostl4,5,6

L. K. Jian7,8 C. J. Davis9,10 C. J. Farrugia1

Received: 10 October 2011 / Accepted: 23 March 2012

Abstract

We present a study of coronal mass ejections (CMEs) which impacted one ofthe STEREO spacecraft between January 2008 and early 2010. We focus ourstudy on 20 CMEs which were observed remotely by the Heliospheric Imagers(HIs) onboard the other STEREO spacecraft up to large heliocentric distances.We compare the predictions of the Fixed- and Harmonic Mean (HM) fittingmethods, which only differ by the assumed geometry of the CME. It is possibleto use these techniques to determine from remote-sensing observations the CMEdirection of propagation, arrival time and final speed which are compared to insitumeasurements. We find evidence that for large viewing angles, the HM fittingmethod predicts the CME direction better. However, this may be due to the factthat only wide CMEs can be successfully observed when the CME propagatesmore than 100 from the observing spacecraft. Overall eight CMEs, originating

1 Space Science Center, University of New Hampshire,Durham, New Hampshire, USA email: [email protected];[email protected] Institute for Astronomy, University of Hawaii, 2680Woodlawn Dr., Honolulu, HI 96822, USA3 University of Rochester, Rochester, NY 14627, USA email:[email protected] Space Science Laboratory - University of California,Berkeley, 94720 CA, USA email:[email protected] Observatory-IGAM, Institute of Physics,University of Graz, Universitatsplatz 5, A-8010, Graz,Austria6Space Research Institute, Austrian Academy of Sciences,Graz 8042, Austria7

Department of Astronomy, University of Maryland, CollegePark, Maryland, USA email: [email protected] Heliophysics Science Division, NASA Goddard SpaceFlight Center, Greenbelt, Maryland, USA9 SFTC Rutherford Appleton Laboratory, Didcot,Oxfordshire OX11 0QX, UK email: [email protected] Department of Meteorology, University of Reading,Berkshire, RG6 7BE

SOLA: N_Lugaz_Sun360.tex; 14 May 2012; 13:00; p. 1

arXiv

:1205.2526v1[astro-ph.SR]11May

2012


2/22

Lugaz et al.

from behind the limb as seen by one of the STEREO spacecraft can be trackedand their arrival time at the other STEREO spacecraft can be successfullypredicted. This includes CMEs, such as the events on 4 December 2009 and9 April 2010, which were viewed 130 away from their direction of propagation.

Therefore, we predict that some Earth-directed CMEs will be observed by theHIs until early 2013, when the separation between Earth and one of the STEREOspacecraft will be similar to the separation of the two STEREO spacecraft in20092010.

Keywords: Coronal Mass Ejections, STEREO, Heliospheric Imagers, Methods

1. Introduction

Remote observations of coronal mass ejections (CMEs) by heliospheric imagers(HIs), combined with in situ measurements provide a unique opportunity tostudy the evolution, propagation and properties of these transients. It also con-

stitutes a main science objective of the Solar-Terrestrial Relations Observatory(STEREO) mission. White-light observations by the HIs onboard the STEREOand by the Solar Mass Ejection Imager (SMEI) onboard Coriolis can be ana-lyzed to determine the direction of propagation of CMEs and their kinematics(e.g. with the methods of Rouillard et al., 2008, Tappin and Howard, 2009,Maloney, Gallagher, and McAteer, 2009, Liu et al., 2010, Lugaz et al., 2010,Lugaz, 2010). These methods can be used in real-time with beacon data fromSTEREO to predict whether or not a CME will hit a spacecraft, and, if a hit isforecasted, to predict the arrival time and speed at 1 AU (Davis et al., 2011).In situ measurements can then be used to test, compare and validate thesepredictions. Another approach is to constrain remote-sensing observations withthe knowledge obtained from in situ measurements (final speed, arrival time).

This is the approach taken, for example, in Woodet al.

(2009), Mostlet al.

(2010), Temmer et al. (2011) and Rollett et al. (2012) to study CME kinematics.Such coordinated studies have been primarily undertaken for Earth-directed

CMEs, which impacted ACE or Wind (Davis et al., 2009; Rouillard et al., 2009;Wood et al., 2009; Mostl et al., 2010; Liu et al., 2010; Liu et al., 2011) butalso for CMEs or corotating interaction regions (CIRs) which impacted VenusExpress (Rouillard et al., 2009; Mostl et al., 2011), MESSENGER or spacecraftorbiting the planet Mars (Williams et al., 2011). Most of the analyses of datafrom planetary missions have been for CMEs which also impacted Earth. Therehave also been a few studies of CMEs which impacted one of the STEREOspacecraft (Wood and Howard, 2009; Mostl et al., 2009).

As the separation between the two STEREO spacecraft increases, stereoscopicviews of a CME into the HI field-of-view are expected to become rarer. Therefore,

stereoscopic methods (Liu et al., 2010; Lugaz et al., 2010) making use of thetwo STEREO views may not be applicable for Earth-directed CMEs from mid-2011 onwards, when the total separation is larger than 180. In addition, it islikely that future missions (e.g., see Gopalswamy et al., 2011) will only haveone spacecraft with a HI instrument. It is therefore important to further testand validate methods using HI observations from a single spacecraft. The two



3/22

Heliospheric Observations of STEREO Events

main fitting methods, the Fixed- (F; Rouillard et al., 2008), and HarmonicMean (HM; Lugaz, 2010), have been compared for three Earth-directed CMEsin Rollett et al. (2012) and for CMEs observed simultaneously by both STEREOspacecraft in 2008 and 2009 (Lugaz, 2010). It was found that the two methods

differ for small (< 30) and large (> 80) viewing angles (Lugaz, 2010; Davieset al., 2012). The largest viewing angle for which these two techniques have beencompared for a Earth-impacting CME is about 67 for the April 2010 CME. Thisspacecraft separation is within the range where the techniques are expected togive similar results. Studying STEREO-directed events has the advantage ofallowing for larger viewing angles compared to Earth-directed CMEs and toprepare for future observations of Earth-directed CMEs in future years.

Here, we analyze the remote-sensing observations of STEREO-impacting CMEsduring the years 20082010. We determine the CME direction of propagation andspeed by fitting the remote-sensing observations with the two fitting techniques(F and HM). The predicted hit/miss, arrival time and speed are then comparedwith in situ measurements. In this way, we are able: i) to assess which of the

two methods is able to better determine some of the CME properties measuredin situ and, ii) to determine how the accuracy of the methods is affected bythe increasing separation of the STEREO spacecraft from the Sun-Earth line.We use STEREO-directed events from 2009 to 2010 to prepare for the largeseparation between the Earth and the STEREO spacecraft which will reach125 by January 2013. This allows us to establish what might be the largestviewing angle for which a CME can be observed into HI-2.

In Section 2, we summarize briefly the two fitting techniques used in thisarticle and explain our data selection process. In Section 3, we present thedetailed analysis of one event, which occurred on 4 December 2009, when theseparation between the two STEREO spacecraft was 130. In Section 4, wepresent the results from the analysis of 20 events which occurred between thebeginning of 2008 and mid-2010. The conclusions of our investigation are drawn

in Section 5.

2. Analysis Techniques and Data Selection

2.1. Fitting Methods

In the field-of-view of a heliospheric imager, the position of a density feature(e.g., a Stream Interaction Region SIR or a CME) is measured as the anglebetween the observing spacecraft, the Sun and the density structure, and it iscommonly referred to as the elongation angle. When CMEs are observed to largeelongation angles, the time-elongation data can be fitted to analytical functionsand a single value for the CME speed and direction of propagation can be derived

(Sheeley et al., 1999). Two common methods are the F (Rouillard et al., 2008)and the HM fittings (Lugaz, 2010). We will focus here on these two methods only.A schematic view of the two methods is shown in the bottom panel of Figure 1.Here, we use the following notation: is the elongation angle, the direction ofpropagation, which is assumed to be fixed, dST the heliocentric position of theobserving spacecraft (STEREO) and t the time.



4/22

Lugaz et al.

Assuming a constant propagation speed, V, and a single plasma element, therelation between elongation and distance of Kahler and Vourlidas (2005) can beinverted to obtain the F fitting relation (Rouillard et al., 2008):

= arctan

V t sin

dST V t cos

. (1)

A measured time-elongation profile can be fitted to a profile of calculatedelongations given by Equation (1). The F approximation assumes that theCME is a single point and therefore, it is not straightforward to determine whenthe fitting results implies a hit or a miss and what is the expected arrival time andspeed of different parts of the CME. Previous works (Davies et al., 2009; Mostlet al., 2011) made the assumption that the F fitting predicts a hit if the best-fitted direction is within 1015 of a spacecraft and that the arrival time andspeed are the same for all parts of the front. Here, we use a looser criterion of20 around the best-fit direction and we also assume that all parts of the CME

front arrive at the same time with the same speed. This criterion of20

isused to include the uncertainty in the best-fit direction of propagation, whichis typically 1015 and the fact that even narrow density features are expectedto have a half-width of about 10. Assuming a half-width wider than 20 is notconsistent with the underlying assumption of the F approximation: the factthat the observed signal originates from the apex of the CME. This can be seenin the bottom panel of Figure 1 for a large viewing angle: if the CME is assumedto be wide, the wings of the CME are at a larger elongation angle than the apexof the CME (see also discussion in Howard and Tappin, 2009).

Another simple assumption for what is observed by HIs, as proposed by Tap-pin and Howard (2009) and Lugaz, Vourlidas, and Roussev (2009), is as follows.The CME front can be modeled as a locally circular front with a diameter equalto the distance of the apex (or equivalently, a front which is anchored at the Sun).It is further assumed that the measured elongation angle simply correspondsto the angle between the Sun-spacecraft line and the line-of-sight tangent tothis circular front. Assuming a constant propagation speed and the geometryexplained above, the HM fitting relation can be obtained (Lugaz, 2010; Liuet al., 2010; Mostl et al., 2011):

= arctan

V t sin

2dST V t cos

+ (2)

arcsin

V t

(2dST V t cos )2

+ (V t sin )2

.

A measured time-elongation profile can also be fitted to profiles of calculatedelongations given by Equation (2). The fitted speed, Vbestfit, and direction,bestfit, correspond to that of the nose (apex) of the CME. To determine thepredicted speed and arrival time at a spacecraft which is not hit directly by thenose of the CME, a correction is needed as derived in Mostl et al. (2011). For a



5/22


spacecraft at a heliocentric distance dspacecraft and separated by an angle withthe observing spacecraft, the predicted speed is Vspacecraft = Vbestfit cos( bestfit) and the predicted time is tarr = dspacecraft/Vspacecraft.

It should be noted that both fitting methods assume a constant propaga-tion speed and a constant direction. Using stereoscopic methods (Liu et al.,2010; Lugaz et al., 2010; Byrne et al., 2010), it is possible to relax these twoassumptions and derive for all pairs of observations the CME direction andposition (and speed). However, it is unlikely that the two STEREO spacecraftcan observe a CME into HI-2 field-of-view when the separation between thespacecraft is 180 or beyond. The technique of Howard and Tappin (2009) can beused to derive the CME position without assuming a constant propagation speed,although it is not straightforward to decouple the effects of the varying speedfrom the change in the CME shape. Due to their simplicity, fitting techniquesare used for real-time space weather forecasting (Davis et al., 2011) and forscientific analyses by a number of groups (Wood and Howard, 2009; Temmeret al., 2011; Rollett et al., 2012) and we focus here on these techniques.

2.2. Data Selection

Following other researchers (e.g., see Richardson and Cane, 2010), we startfrom ICMEs measured in situ at 1 AU by one of the two STEREO spacecraftand attempt to identify, first, their heliospheric sources as measured in the HIinstruments and, second, their coronal sources. Our starting point is the list ofICMEs observed by IMPACT (Luhmann et al., 2008) and PLASTIC (Galvinet al., 2008) onboard STEREO as maintained at www-ssc.igpp.ucla.edu/jlan/STEREO/Level3. The determination of the ICMEs and their characteristics canbe found in Jian et al. (2006). We focus on ICMEs detected from January 2008 toJuly 2010: 21 by STEREO-A and 26 by STEREO-B. For this list of 47 ICMEs,we attempt to find their heliospheric counterparts in remote-sensing images.We use the list summarized on the Rutherford Appleton Laboratory (RAL)website www.stereo.rl.ac.uk/HIEventList.html of heliospheric transients observedby the HIs (Eyles et al., 2009) part of SECCHI (Howard et al., 2008) onboardSTEREO. This list consists of the time-elongation tracks measured on J-maps(Davies et al., 2009) along elevation 0 for various transients. Some of thetime-elongation measurements from this list have been used in previous studies(Davis, Kennedy, and Davies, 2010; Lugaz, 2010). On the RAL website, thetime-elongation datapoints for each track are plotted; in the present study, weuse the corresponding time-elongation measurements.

J-maps are constructed using running differences of SECCHI images; intensityenhancements with respect to the previous image appear as bright regions whiledepletions appear as black regions (Davies et al., 2009). As noted in Davis,

Kennedy, and Davies (2010), the clearest feature to track is the black front,i.e. the transition between the bright regions and the dark regions (maximumcontrast). This black front is typically 46 hours behind the start of the CMEand, therefore, our tracking introduces a systematic offset in the predicted arrivaltime of an event. A J-map and an example of the tracking of a black front(with red circles) are shown in the top panel of Figure 1. In Lugaz et al. (2010),



6/22

Lugaz et al.

01 02 03 04 05 06 07 08 09 10 11Day in December

10

20

30

40

50

Elongation

(o)

Figure 1. Top and Middle: Analysis of the 4 December 2009 CME observed by STEREO-B.Top: J-map and datapoints (red circle) corresponding to the black front. Middle: Datapoints(plus symbols) and best-fit solution obtained with HM fitting method (solid line) and F fittingmethod (dash-dot line) fittings. In this panel, beta refers to the best-fit direction with respectto the Sun-Earth line. Bottom: Sketch of the Fixed- (orange) and Harmonic Mean (green)approximations.

we compared the direction of propagation and distance derived by stereoscopicmethods for the 2008 April 26 CME (CME # 2 in the present study) using

the bright front and the black front of the CME. We found that in the HIfields-of-view the kinematics and direction of the CME using the two tracks arenearly identical. As a summary, in order to follow the CMEs to large elongationangles, we use the tracks of the black front of the CMEs; this introduces asystematic offset in the predicted arrival time but it is not expected to changethe predicted arrival speed and CME direction. Since no study has yet focused



7/22


on determining the exact value for this offset, we do not correct the predictedarrival times for it.

For each ICME from the IMPACT list, we have looked in the RAL list fortransients whose starting time is within a window of 36 to 120 hours prior

to the arrival time of the ICME. We use HI-1 movies to identify the CMEsand to associate them with one (or multiple) of the tracks from the RAL list.In most cases, there is only one CME for which the expected arrival time iswithin 24 hours of the actual arrival time and whose time-elongation profileshows the required trend (deceleration for large viewing angles, acceleration forsmall ones). However, there are often multiple tracks associated with each CME,corresponding, for example, to the two dense fronts which bracket the CMEs (see,e.g., Savani et al., 2011) and sometimes to core material. In these cases, we havechosen the track corresponding to the leading edge of the CME. Sometimes,there are no associated tracks, either because no CME was visible in the field-of-view of the HIs or because there is no track in the RAL list associated with agiven CME (often because the CME is faint and not visible to large elongationangles, sometimes because of data gaps). Overall, we find 20 ICMEs ( 40%)for which there is a corresponding track in the RAL list. Using information fromthe HI movies, we identify the CMEs in COR-2 and COR-1, when possible. The20 CMEs as well as their first appearance in COR-1 or COR-2 are summarizedin Table 1. Some of the CMEs are streamer blowouts, which typically form atheights of 23 R, start at low speed and may not be associated with any clearsignature in the low corona (e.g., see Robbrecht, Patsourakos, and Vourlidas,2009). For these CMEs, we do not indicate a starting time but only a startingday in Table 1. The information from COR-1 and COR-2 is not directly usedin this study, but it is listed for completeness and, for future references, if ourlist is to be used in future studies. In Table 1, CMEs are ordered by the totalseparation between the two STEREO spacecraft. Since each CME was observedremotely by one STEREO spacecraft and measured in situby the other, the total

separation gives an estimate of the viewing angle at which STEREO remotelyobserved the CME.

The proportion of ICMEs for which we find remote-sensing observations intothe outer heliosphere allowing a successful prediction decreases with increasingseparation between the STEREO spacecraft, from 10/14 in 2008 to 9/21 in 2009and only 1/12 in 2010. However, there are 5 CMEs which were tracked into HI-2field-of-view with a spacecraft separation of more than 120.

3. Case Study: 4 December 2009 CME Event

In this section, we give an overview of the analysis of these 20 CMEs by study-ing in detail one CME observed remotely more than 30 behind the limb. We

compare the two analysis techniques and discuss the implications of the CMEdetection and tracking for CME geometry and the Thomson scattering.

3.1. Observations and Fitting

In early December 2009, STEREO-A and B were separated by almost 130

(STA = 63.6, STB = 65.8

) with STEREO-A at 0.97 AU and STEREO-B



8/22

Lugaz et al.

Table 1. STEREO-impacting and STEREO-observed CME list: The columns

show the CME number in this study, the ICME number (with year) from theIMPACT list, the track number from the RAL list, the date of first detectionin COR-1 or COR-2 and the separation between STEREO-A and STEREO-Bfrom left to right. (f) indicates a filament eruption.

CME IMPACT RAL 1st Detection Tot. Separation

2008

1 7B 327A 31 Jan. (COR-2B) 45.3

2 9B 399A 26 Apr. 14:25 (COR-1B) 49.8

3 2A 172B 7 May (COR-2A) 51

4 11B 428A 1 Jun. 10:25 (COR-1B) 54.8

5 3A 214B 30 Jun. (COR-2A) 58.3

6 12B 514A 10 Aug. (COR-2B) 67.2

7 4A 247B 30 Aug. 16:05 (COR-1A) 71.3

8 5A 270B 26 Oct. (COR-1A) 81.3

9 6A 290B 23 Nov. 11:25 (COR-1A) 85.2

10 14B 595A 27 Dec. 02:25 (COR-1B) 88.4

2009

11 10B 611A 8 Jan. 4:00 (COR-1B) 89.3

12 1A 328B 21 Jan. 16:45 (COR-1A, f?) 90

13 12B 762A 10 Jul. (COR-2B) 104.1

14 13B 771A 26 Jul. 11:35 (COR-2B) 106.5

15 18B 818A 26 Sep. 17:45 (COR-1B, f ) 118.4

16 19B 858A 5 Nov. 8:05 (COR-1B, f) 125.3

17 7A 466B 8 Nov. 5:25 (COR-1A,f) 125.8

18 20B 881A 22 Nov. (COR-2B) 127.9

19 9A 489B 4 Dec. 6:50 (COR-1A, f) 129.5

2010

20 3A 575B 19 Apr. 14:25 (COR-1A,f) 139.6

at 1.06 AU. There was a filament visible in the SECCHI/Extreme UltravioletImager (EUVI) 304 A behind the limb of STEREO-B and disk centered as seenby STEREO-A. It started to rise at around 02:00UT on 4 December and eruptedstarting at around 06:45UT. It was first visible in COR-1B at 06:50UT and inSOHO/C2 at 08:50 and appeared as a weak halo in COR-1A. It entered intoCOR-2B field-of-view at 11:55, C3 at 12:20 and HI-1B at 20:49. We show a J-map of this CME in the top panel of Figure 1. Using this J-map, it is possibleto track the CME for a total of 85 hours and up to an elongation angle of about33 (26 datapoints, shown with red circles in the top panel of Figure 1). While

erupting from S50, the filament and associated CME were significantly deflectedin the low corona towards the equator and the CME enters COR-2B field-of-view propagating at a position angle close to 260265. As shown in Figure 2, itappears to propagate close to PA 270 well into HI-1 field-of-view.

We fit the data from STEREO-B/SECCHI with the two fitting methods.The F fitting yields a best-fit speed of 338 63 km s1 and a direction of



9/22


Figure 2. STEREO remote-sensing observations of the 4 December 2009 CME: EU-VI-A-304 A at 05:56:15 UT, COR-1B at 09:26 UT with EUVI-B-304 A at 09:17:04 UT inlaid,COR2-B at 17:09 U and HI1-B at 06:49 UT on 5 December, from top left to bottom right. Inthe EUVI-A image, the two ends of the filament are indicated with the black arrows.

30.2 14.5 with respect to the Sun-Earth line. The HM fitting of the samedata yields a best-fit speed for the nose of 401 48.5 km s1 and a direction of62.214. The derived CME directions, with respect to the observing spacecraft(STEREO-B), are 96 (F) and 128 (HM). The middle panel of Figure 1 showsthe measured datapoints and the best-fit time-elongation track from the twomethods. It can be seen that, although both methods fit the data equally well,

there is a difference of more than 30

in the predicted direction of propagation ofthe CME. As expected from previous theoretical and statistical analyses (Lugaz,2010; Davies et al., 2012), for this CME observed behind the limb by STEREO-B,the HM fitting gives a larger angle than the F fitting method.

The predicted direction based on the HM method is almost directly towardsSTEREO-A, consistent with a predicted direct hit at the spacecraft. The pre-



10/22

Lugaz et al.

dicted arrival speed is 401 48.5 km s1 with a predicted arrival time of 18:20UT on 8 December. Based on the F fitting method, the CME propagates morethan 30 off STEREO-A (towards Earth), consistent with a predicted miss.If the CME were to hit STEREO-A, the predicted speed would be 338 63km s1 and the predicted arrival time 12:30 UT on 9 December. As indicated inFigure 3, a fast forward shock was detected in situ by STEREO-A at 23:38 UTon 8 December. A magnetic cloud was detected from 9 December at 09:00 UTto 10 December at 23:13 UT. The speed of the dense sheath behind the shockwas about 330-340 km s1. The detection of a shock and a magnetic cloudappears to validate the HM fitting which predicts a direct hit at STEREO-A.In addition, the error in the arrival time is slightly better for the HM fittingmethod as compared to the F fitting method (about 5 hours for the HM fitversus 13 hours for the F fit) but the velocity is better predicted by the Ffitting method (see Figure 3). However, for this CME the F method predicts amiss or, at best, a glancing blow. In fact, it is not consistent to assume that theposition is given by the F approximation and that part of CME 30 away from

the CME nose arrives at the same time and with the same speed at 1 AU asthe nose of the CME. Therefore, the arrival time and arrival speed predictionsderived using the F approximation should be considered with extreme cautionfor this case study.

Additionally, as indicated in Figure 3, there appear to be two distinct mag-netic clouds (highlighted in blue and noted as MC1 and MC2) measured in situ.The first magnetic cloud, with a low inclination ( 10 with respect to the solarequatorial plane) is the better candidate to correspond to the low-inclinationfilament observed in EUVI. We have not been able to find a clear heliospheric(or solar) source of the second, higher inclination ( 60) magnetic cloud. Whilethe presence of two clouds has no direct influence on our study which focuses onthe arrival time and speed of the density structure ahead of the cloud(s), thisexample illustrates that many events are not isolated CMEs.

3.2. Further Analysis and Consequences of the Observations for CME Geometry

While the direction of propagation of the CME seems to be well reproducedwith the HM fitting, the predicted arrival speed and arrival time are off. Inthis respect, it is interesting to investigate in more detail the kinematics of theeruption. To do so, we follow Wood and Howard (2009), Temmer et al. (2011)and Rollett et al. (2012) by relaxing the assumption of constant speed. We stillassume that the CME propagates radially outward with the direction given bythe HM fitting method (62 14). While deflection is frequent in the corona,prior studies have shown that the assumption of radial propagation holds in theheliosphere (Lugaz et al., 2010). In addition, we take an a posteriori approach

similar to that of Rollett et al. (2012) to constrain the CME parameters withknowledge of the actual arrival speed and arrival time of the CME.

We first focus on matching the arrival speed, measured in situ to be about330340 km s1. In Figure 4, we show the kinematics of the CME assuming aconstant direction of propagation but not a constant speed. The three curvescorrespond to the best-fit direction of 62 (black) and the lower and upper



11/22


10

0

10

B(nT)

STEREOA IMPACT/ PLASTIC December 89 2009 ICME (RTN coordinates)

|B|B

R

BT

10

0

10

BN

(nT)

shock sheath MC1

MC2

250

300

350

400HM fits

FP fits

Vp(km/s)

010203040506070

Np(cm

3)

0.01

0.1

1

Tp(MK)

Tp

Texp

0.01

0.1

1

10

p

00:00 12:00 00:00 12:00 00:00 12:00 00:00 12:00 00:00 12:00

0.02

0.04

0.06

0.08

Ptot

(nP)

Ptot

PB

Pp

12/08 12/09 12/10 12/11 12/12

Figure 3. In situ measurements by STEREO-A IMPACT and PLASTIC of the ICME on9 December 2009 showing the magnetic field (top 2 panels), proton velocity, density andtemperature, the plasma beta and total pressure. The two vertical dashed lines show thepredicted arrival time of the front using the HM (blue) and F fitting (red) methods. The twosolid lines show the expected speed and typical error bars of 5 hours for the arrival time forthe two methods. MC1 and MC2 are the two magnetic clouds measured at 1 AU, see text fordetails.

bounds of the 1-sigma interval 48 (red) and 76 (green). In all three cases, a hitat STEREO-A is predicted since the direction of propagation is within 20 ofthe spacecraft position. It can be seen from the top panel of Figure 4 that themeasured speed can be reproduced using the time-elongation data for a CME

propagating between 48

and 62

from the Sun-Earth line.We then focus on matching the measured shock arrival time by varying the

CME direction of propagation. To do so, we assume, after the last data point,a constant CME speed equal to the average speed of the front over the last 24hours of observations. For the best-fit direction, the arrival time at STEREO-A under these assumptions is expected to be 19:00 UT on 8 December. The



12/22

Lugaz et al.

exact arrival time is obtained for a direction of 58 corresponding to a speed atSTEREO-A of 375 km s1. In the J-map, for better contrast, the black front(about 6 hours behind the shock) is being tracked (see top panel of Figure 1).Assuming the arrival time corresponds to that of this black front, the arrivaltime can be matched with a direction of propagation of 54 and the final speedat STEREO-A is 350 km s1 in good agreement with the speed measured in situ.Overall, taking an a posteriori approach similar to Rollett et al. (2012), we findthat the measured arrival time and arrival speed can be reproduced assuming(i) a CME geometry given by the HM approximation of Lugaz, Vourlidas, andRoussev (2009) and (ii) a constant direction of propagation between 50 and 60

with respect to the Sun-Earth line (or about 10 East of STEREO-A).From this analysis, there is strong evidence that the CME did not deflect

much in the longitudinal direction and remained directed towards STEREO-A.Therefore, STEREO-B/SECCHI was able to image to large elongation angles aCME propagating with a direction of 120 with respect to the Sun-spacecraftline. In addition, the time-height data derived using the HM approximation is

consistent with the observed arrival time and the the measured speed in thesheath.

One of the limitations of the HM and F fitting models is that they neglectthe effect of Thomson scattering. If the elongation angle corresponds to theintersection of the CME front with the Thomson sphere rather than the tangentto the CME front, as assumed by the HM model, it is still possible for a CME tohave a shape which would give the same arrival time and the same kinematics.Such a geometry is shown in green in the bottom panel of Figure 4 next tothe assumption of the HM method in pink. The main difference between thetwo assumptions is about the width of the CME. If we assume that what isobserved is the intersection of the CME front with the Thomson sphere, the CMEfront needs to be significantly wider than if we use the HM approximation. Inaddition, the signal comes from the wings of the CME in order to still resultin a hit at STEREO-A. It is likely that the real geometry of wide CMEs issomewhere between these two descriptions, although it should be noted that theCME density structure needs to be as wide as 160 for it to intersect with theThomson sphere.

4. Comparison of the Two Fitting Methods

The full results of our analysis are shown in Table 2 and top panel of Figure 5for the CME speed and direction, Table 3 and bottom panels of Figure 5 for thepredicted arrival time, hit/miss and the predicted and measured speeds at 1 AU.An online version of the three tables can be found on the electronic supplement

to the online article. For the arrival time, we list the shock arrival time, whena shock is present and the start of the ICME when there is no shock. Previousstudies have shown that the bright structure observed in HI-2 corresponds to thesheath ahead of the magnetic cloud (Wood et al., 2009; Mostl et al., 2009; Mostlet al., 2011; Liu et al., 2010; DeForest, Howard, and Tappin, 2011). Therefore,for fast CMEs, the shock arrival time, corresponding to the beginning of the



13/22


Figure 4. Top: Velocity of the apex of the 4 December 2009 CME assuming that (i) thedistance is given by the HM method, (ii) the CME is propagating radially outward at a angleof 48 (red), 62 (black) and 76 (green) from the Sun-Earth line. Assuming precision of 0.2

and 0.5in HI-1 and HI-2, respectively the error bars in the velocity are about 70100 km s1.Bottom: Sketch of the geometry of the 4 December 2009 CME assuming a viewing angle of 120.The Thomson sphere of STEREO-B is shown in dashed blue. The pink front corresponds to theHM assumption and the green front would give the same kinematics but the signal originatesfrom the intersection of the front with the Thomson sphere.

sheath region, is the best time to compare to the predicted arrival time from the

fitting methods. For slow CMEs which do not drive a shock, we use the startingtime of the ICME. As for the measured speed of the CME, following Jian et al.

(2006), we report the maximum speed of the ICME, which is usually reached at

the start of the ICME (due to the typical decreasing speed profile inside ICMEs)

and it usually corresponds to the speed in the sheath. The uncertainty in the

arrival time due to the uncertainty in the best-fit speed is typically 57 hours



14/22

Lugaz et al.

for the F method and 810 hours for the HM method. However, for mostof the CMEs observed in 2009, the uncertainty in the arrival time is 1 dayfor the following reason. For the CMEs in our studies observed in 2009, whenthe viewing angle is 90 or more, time-elongation tracks tend to be shorter. Thisis because a CME propagating behind the limb observed at 1 AU is only at anelongation angle of 35-45, most CMEs in 2009 from our study are only observedup to 35. Previous studies (e.g., see Mostl et al., 2011) have shown that time-elongation tracks must extend well beyond 30 for the uncertainty in the arrivaltime to be only of the order of 6 hours.

In the remaining part of the article, we compare the results of the two methodsto determine how well they are able to reproduce in situmeasurements. The firstcriterion under which we judge the fitting methods is whether or not they areable to successfully predict a hit. Following the discussion in section 2.1, weconsider that a fitting method predicts a hit if the best-fit propagation angle iswithin 20 of the Sun-spacecraft line. In a later stage, we consider the predictedarrival time and final speed as compared to in situ measurements. Predicting

these quantities from remote-sensing observations is only important if the fittingmethods can correctly identify a CME which is set to hit a spacecraft. In Figure 5,the data points are color-coded with the total spacecraft angular separation frompurple for an angle of 45 (Jan. 2008) to dark red for an angle of 140 (Apr.2010). Square symbols represent events for which both fitting methods predicta hit, triangle symbols correspond to events for which the F fit predicts a hitbut the HM fit predicts a miss and circular symbols to the opposite case. Below,we discuss the different cases.

4.1. Events Successfully Predicted by Both Methods

Nine out of the ten events in 2008 are successfully predicted by both methods(the exception is CME 5) but the same is true for only three out of the ten eventsin 20092010 (first 3 events). For nine out of these 12 events, the arrival timepredicted from the two fitting methods is nearly identical (see square symbols inthe bottom panels of Figure 5). The exceptions are events 10, 11 and 13 (all forseparation greater than 85). This finding is also confirmed by Earth-directedevents, for which both methods have essentially given the same predictions upuntil the beginning of 2011. The arrival time of events 11 and 13 is significantlybetter reproduced with the HM fitting method than with the F method. Thiscomes from the fact that the HM best-fit velocity is usually higher than thebest-fit velocity from the F method (see also Davies et al., 2012). It is alsoconsistent with the fact that for viewing angles greater than 80, the HM fittingmethod is expected to give better results than the F method (Lugaz, 2010).Event 10 is different because both methods predict a CME direction more than

25 away from the spacecraft (towards Earth), i.e. both methods predict a miss.However, this event was both observed in situ by ACE and STEREO-B, so weconsider that the best-fit angle gives an accurate prediction.

As noted before, the back of the white front in the J-map (black front) isbeing tracked. We consider that this front is 46 hours behind the CME frontas seen in J-maps. Therefore, there is a bias of the two methods towards finding



15/22


Table 2. Best-fit propagation speed and direction with respect to the Sun-Earth Line ofthe 20 CMEs and separation angle of the spacecraft hit by the CMEs. The separation angleis the one between the impacted spacecraft and Earth and it is therefore different from theone listed in Table 1 between STEREO-A and STEREO-B.

CME F Speed F Direction HM Speed HM Direction In situ Separation(km s1) () (km s1) () ()

1 304 13.5 28.2 14.5 316 17.5 30.2 25.5 23.6

2 544 44 33.2 18 548 30.5 33.2 30.5 25.8

3 368 17 34.7 12.5 373 27 36.7 27 24.3

4 371 13 23.8 5.5 379 27.5 33.8 15 29.2

5 314 5.5 11.8 6.5 330 13 0.2 17 27.2

6 325 17 24.8 11 329 18.5 30.8 16 36.2

7 350 8.5 33.7 10.5 353 14.5 36.7 27 33.3

8 358 40 38.9 17.5 369 78 52.9 37 40.1

9 411 111 32.8 26.5 436 168 49.8 49.5 43.2

10 425 56.5 11.4 17.5 431 37.5 19.4 16.5 42.6

11 351 62.5 33.2 17 379 35 55.2 35 46.6

12 376 12.5 20.9 6.5 386 41 33.9 20.5 4313 286 19 46 8 323 58 74 12.5 55.2

14 407 114 39.4 17.5 505 174 75.4 27 50.1

15 454 205 64.6 27 699 201 113.6 109 57

16 283 138 56.6 21 455 198 107.6 32.5 64.4

17 339 61 32.3 15 390 66 61.3 20.5 63.1

18 269 26 24.5 52 306 66.5 52.5 20.5 64.4

19 338 63 30.2 14.5 401 48.5 62.2 14 63.7

20 358 59.5 24.3 11 439 137 59.3 22.5 68.9

later arrival times. In fact, if we look at the average (signed) arrival time error, itis +6 hours for the HM and +10.6 hours for F fitting methods. Only five eventsare predicted early by one or both methods. We believe part of this average errorin the arrival time is directly attributable to the fact that we tracked the blackfront in the J-map.

4.2. Hits Predicted by the F but not HM Fits

For CMEs 5, 14, 15 and 16, the best-fit CME direction from the F fit is within20 of the STEREO spacecraft but the best-fit direction from the HM is morethan 20 away. CME 5 is a very faint event for which the best-fit angle fromboth methods indicate an expected hit at L1, which was not recorded. It is likelythat the assumption of wide CME used in the HM fitting method does not apply

for this CME. CME 14 is a relatively fast event (460 km s1

at 1 AU from thein situ measurements) observed at a viewing angle of about 100. For CMEsfaster than the solar wind, the assumption of constant propagation speed usedin the fitting methods results in an additional error: the physical decelerationdue to the interaction with the solar wind (Cargill et al., 2000; Vrsnak et al.,2010; Davis, Kennedy, and Davies, 2010) is fitted as a geometrical deceleration



16/22

Lugaz et al.

Table 3. CME number, predicted hit (H) or miss (M), predicted arrival times (UT) and arrival speeds (km s1)for the F and HM methods for columns 1 to 7, and actual arrival time and arrival speed and impacted spacecraftfor columns 8 to 10. For CME # 10, both methods predict a direction between STEREO-B and ACE at Earthand the CME impacted both spacecraft, which we note as H* (see text for details). The arrival times (measuredand predicted) are rounded to the nearest ten minutes.

CME H/M F tarr F V H/M HM tarr HM V tarr V STEREO

2008

1 H 6 Feb 15:50 304 H 6 Feb 14:50 314 5 Feb 20:30 385 B

2 H 29 Apr 19:30 544 H 29 Apr 20:25 541 29 Apr 14:10 490 B

3 H 11 May 11:25 368 H 11 May 12:30 367 11 May 06:30 350 A

4 H 7 Jun 01:30 371 H 7 Jun 01:30 375 6 Jun 15:40 430 B

5 H 5 Jul 11:20 314 M 5 Jul 18:30 282 5 Jul 00:50 360 A

6 H 16 Aug 05:50 325 H 16 Aug 04:50 329 15 Aug 12:00 365 B

7 H 4 Sep 22:00 350 H 4 Sep 21:30 353 4 Sep 13:20 385 A

8 H 31 Oct 16:00 358 H 31 Oct 15:20 361 31 Oct 12:20 400 A

9 H 28 Nov 16:50 411 H 28 Nov 12:30 432 28 Nov 21:50 380 A

10 H* 31 Dec 10:20 425 H* 31 Dec 21:30 386 31 Dec 02:00 460 B

2009

11 H 14 Jan 01:40 351 H 13 Jan 18:40 375 13 Jan 5:20 400 B

12 H 26 Jan 07:50 376 H 26 Jan 07:10 381 25 Jan 18:20 400 A

13 H 17 Jul 06:10 286 H 16 Jul 20:30 306 16 Jul 17:10 330 B

14 H 31 Jul 00:00 407 M 30 Jul 12:40 456 31 Jul 02:20 460 B

15 H 1 Oct 09:40 454 M 2 Oct 03:30 385 2 Oct 17:20 360 B

16 H 12 Nov 01:50 283 M 12 Nov 01:50 332 10 Nov 18:50 370 B

17 M 14 Nov 10:05 339 H 13 Nov 19:00 390 14 Nov 08:00 340 A

18 M 28 Nov 18:10 269 H 28 Nov 02:10 345 27 Nov 12:40 400 B

19 M 9 Dec 12:30 338 H 8 Dec 18:20 401 8 Dec 23:40 350 A

2010

20 M 24 Apr 7:30 358 H 23 Apr 13:00 433 23 Apr 00:35 450 A

associated with a large viewing angle. This effect is particularly pronounced forthe HM fitting method (Lugaz and Kintner, 2012) because it has, intrinsically,larger errors than the F fitting method.

For CMEs 15 and 16, the HM fits give a better match for the arrival time thanthe F fits. Event 15 corresponds to a small track due to a data gap starting ataround 22:00 UT on 28 September (when the CME is around 15 elongation).Previous works have shown that density transients must be imaged to muchlarger elongations to be fitted with accuracy (Williams et al., 2009; Mostl et al.,2011). The filament eruption originated from the eastern side of the solar disk

as seen from STEREO-B, corresponding to an angle of propagation likely to begreater than 130. Event 16 is another filament eruption from behind the easternlimb as seen in STEREO-A, but it is unclear from STEREO-B images whereit originated from. In both events, the HM method is better mostly due to thecorrection in the arrival time of Mostl et al. (2011). However, the best-fit anglesobtained by the HM fitting method are unrealistically large ( 170 viewing



17/22


Figure 5. Top: Predicted propagation angle with respect of the hit spacecraft (x-axis: Ffitting method, y-axis: HM fitting method). Bottom left: Same as top but for the error inthe arrival time. Bottom right: Same as left but for the absolute value of the error in arrivaltime. In all panels, the symbols are color-coded with the total spacecraft separation (purple:

45

, dark red: 140

). In the bottom panels, square symbols are for CME events for whichboth methods would have predicted a hit, triangle those for which only the F fitting methodwould have predicted a hit and circle those for which only the HM fitting method would havepredicted a hit.

angle). The fact that HM fitting method yields a better arrival time estimatethan the F fitting method is probably just coincidental.

4.3. Hits Predicted by the HM but not F Fits

For the last four events of our list (17, 18, 19 and 20), a hit was only predictedwith the HM fit. For the last three events (18, 19, 20), the arrival time is bestpredicted with the HM fitting method as well, whereas for event 17, the error is

smaller using the F fit. However, for event 17, the CME direction of propagationpredicted by the F fitting is more than 30 away from the spacecraft, whichmakes it unlikely that the F fitting predicts a hit. The filament eruption forevent 17 is seen almost as disk-centered from STEREO-A, which correspondsto the direction predicted by the HM fit. The larger error originates from thelarger speed as compared with the F fit.



18/22

Lugaz et al.

5. Discussion and Conclusions

In this article, we have examined STEREO-impacting CMEs, combined in situmeasurements with remote observations by STEREO/SECCHI and compared

the prediction for hit/miss, arrival time and speed for the two most commonlyused fitting methods. We have started from a list of 47 ICMEs observed usingPLASTIC and IMPACT from January 2008 to June 2010, during which timethe separation between the two STEREO spacecraft increased from 45 to 140.We have been able to identify the remote-sensing observations for 20 ICMEs( 40%). For some ICMEs for which we were unable to identify a counterpart inthe remote-sensing observation, this was due to data gaps or poor observationsleading to short tracks. It is also likely that some of the ICMEs identified at1 AU were small flux ropes embedded in the solar wind or in stream interactionregions (SIRs). This type of ICMEs is unlikely to be observed behind the limb.

The proportion of ICMEs for which we find remote-sensing observations de-creases with increasing viewing angle from 10/14 in 2008 (separation less than

90

) to only 1/12 in 2010 (separation greater than 130

). However, we find sixCMEs viewed more than 20 behind the limb, which can be tracked into HI-2field-of-view. Five of these six CMEs were filament eruptions, hence, associatedwith wide CMEs. Therefore, we predict that some Earth-directed CMEs will besuccessfully tracked by SECCHI until early 2013, when the STEREOEarth sep-aration will be comparable to the STEREO-ASTEREO-B separation in early2010. It should be noted that the first CME observed by SECCHI propagated2030 behind the limb (Lugaz et al., 2009) and was imaged to 40 elongationangle. If the measured signal in the HI instruments come from the part of theCME closest to the Thomson sphere, only very wide CMEs might be observedup to these large viewing angles (similar to the original argument of Vourlidasand Howard, 2006). If the Thomson scattering effects are relatively negligibledue to the extended and non-uniform density structure of the CME front, thenthe detected signal is likely to originate from the tangent to the CME frontas proposed by Tappin and Howard (2009) and Lugaz, Vourlidas, and Roussev(2009). In this case, less wide ( 60) Earth-directed CMEs might be observedin 2012.

We have compared the fitting method based on the Fixed- approximation(Sheeley et al., 1999), where the CME apex is assumed to be observed at alltimes, with that based on the Harmonic Mean approximation (Lugaz, Vourlidas,and Roussev, 2009), where the tangent to a circular CME front is assumed tobe observed. We confirm that both methods give nearly identical results for thepropagation angle of CMEs viewed less than 80 away from the observing space-craft. We have also found that this holds true for the predicted arrival time andfinal speed of CMEs. We show evidence, based on 20 ICMES, that these methods

provide direction and speed consistent with in situ measurements. For viewingangles larger than 90, we find significant differences between both methods.We find some anecdotal evidence that the HM fitting method might be betteradapted to very large viewing angles as compared to the F fitting method: thearrival time predicted using the HM fit is better (4/8) or identical (2/8) than thearrival time predicted from the F fit in six out of eight CMEs observed behind



19/22


the limb. However, for some fast CMEs, the HM fitting method gives unrealisticresults for the CME speed and direction, for the following reason. The constantvelocity assumption of these methods causes CME deceleration (acceleration)to be interpreted as geometrical deceleration (acceleration), resulting in errorsin the best-fit direction of propagation. This effect has been found to be morepronounced for the HM fitting method as compared to the F method (Lugazand Kintner, 2012). These results should be confirmed in further studies usingEarth-directed CMEs as well as CMEs which impacted planetary missions suchas those dedicated to Mercury, Venus and Mars. In addition, further studiesshould look into more details at the in situ characteristics (speed, duration,strength, orientation, etc...) of ICMEs which were identified in remote-sensingobservations as compared to those for which we could not find a correspondingtrack in J-maps.

Acknowledgements The authors would like to thank the anonymous referee for useful

comments and suggestions, which helped improve the clarity of the manuscript. N. L. was

supported during this work by NSF grants AGS-0819653, AGS-1239699 and AGS-1239704 andNASA grants NNX-07AC13G, NNX-08AQ16G and NNX-12AB28G. P. K. performed research

for this work under a Research Experience for Undergraduates (REU) position at the University

of Hawaiis Institute for Astronomy and funded by NSF. L.K.J.s work was funded by NASAs

SMD as part of the STEREO project, including the IMPACT investigation. This work was also

partially supported by NASA STEREO program through grant NAS5-03131 to UC Berkeley

and UNH and has received funding from the European Union Seventh Framework Programme

(FP7/2007-2013) under grant agreement 263252 [COMESEP]. C. M. was supported by a

Marie Curie International Outgoing Fellowship within the 7th European Community Frame-

work Programme (PIOF-GA-2010-272768 [WILISCME]). SOHO and STEREO are projects of

international cooperation between ESA and NASA. The SECCHI data are produced by an in-

ternational consortium of Naval Research Laboratory, Lockheed Martin Solar and Astrophysics

Lab, and NASA Goddard Space Flight Center (USA), Rutherford Appleton Laboratory, andUniversity of Birmingham (UK), Max-Planck-Institut fur Sonnensystemforschung (Germany),

Centre Spatiale de Liege (Belgium), Institut dOptique Theorique et Appliquee, and Institut

dAstrophysique Spatiale (France).

References

Byrne, J.P., Maloney, S.A., McAteer, R.T.J., Refojo, J.M., Gallagher, P.T.: 2010, NatureCommunications 1, 74. doi:10.1038/ncomms1077.

Cargill, P.J., Schmidt, J., Spicer, D.S., Zalesak, S.T.: 2000, J. Geophys. Res. 105, 7509.doi:10.1029/1999JA900479.

Davies, J.A., Harrison, R.A., Rouillard, A.P., Sheeley, N.R., Perry, C.H., Bewsher, D., Davis,C.J., Eyles, C.J., Crothers, S.R., Brown, D.S.: 2009, Geophys. Res. Lett. 36, 02102.

doi:10.1029/2008GL036182

.Davies, J.A., Harrison, R.A., Perry, C.H., Mostl, C., Lugaz, N., Rollett, T., Crothers, S.R.,Davis, C.J., Temmer, M., Savani, N.P.: 2012, Astrophys. J. 750, 23. doi: 10.1088/0004-637X/750/1/23

Davis, C.J., Kennedy, J., Davies, J.A.: 2010, Solar Phys. 263, 209, doi:10.1007/s11207-010-9535-2.

Davis, C.J., Davies, J.A., Lockwood, M., Rouillard, A.P., Eyles, C.J., Harrison, R.A.: 2009,Geophys. Res. Lett. 36, 08102. doi:10.1029/2009GL038021.



20/22

Lugaz et al.

Davis, C.J., de Koning, C.A., Davies, J.A., et al.: 2011, Space Weather 90, 01005.doi:10.1029/2010SW000620.

DeForest, C.E., Howard, T.A., Tappin, S.J.: 2011, Astrophys. J. 738, 103. doi:10.1088/0004-637X/738/1/103.

Eyles, C.J., Harrison, R.A., Davis, C.J., et al.: 2009, Solar Phys. 254, 387. doi:10.1007/s11207-

008-9299-0.Galvin, A.B., Kistler, L.M., Popecki, M.A., et al.: 2008, Space Sci. Rev. 136, 437.

doi:10.1007/s11214-007-9296-x.Gopalswamy, N., Davila, J.M., St. Cyr, O.C., Sittler, E.C., Auchere, F., Duvall, T.L., Hoek-

sema, J.T., Maksimovic, M., MacDowall, R.J., Szabo, A., Collier, M.R.: 2011, J. Atmos.Solar-Terr. Phys. 73, 658. doi:10.1016/j.jastp.2011.01.013.

Howard, R.A., Moses, J.D., Vourlidas, A., et al.: 2008, Space Sci. Rev. 136, 67.doi:10.1007/s11214-008-9341-4.

Howard, T.A., Tappin, S.J.: 2009, Space Sci. Rev. 147, 31. doi:10.1007/s11214-009-9542-5.Jian, L., Russell, C.T., Luhmann, J.G., Skoug, R.M.: 2006, Solar Phys. 239, 393.

doi:10.1007/s11207-006-0133-2.Kahler, S.W., Vourlidas, A.: 2005, J. Geophys. Res. 110, 12. doi:10.1029/2005JA011073.Liu, Y., Thernisien, A., Luhmann, J.G., Vourlidas, A., Davies, J.A., Lin, R.P., Bale, S.D.:

2010, Astrophys. J. 722, 1762. doi:10.1088/0004-637X/722/2/1762.Liu, Y., Luhmann, J.G., Bale, S.D., Lin, R.P.: 2011, Astrophys. J. 734, 84. doi:10.1088/0004-

637X/734/2/84.

Lugaz, N.: 2010, Solar Phys. 267, 411. doi:10.1007/s11207-010-9654-9.Lugaz, N., Kintner, P.: 2012, Solar Phys., in press. doi: 10.1007/s11207-012-9948-1Lugaz, N., Vourlidas, A., Roussev, I.I.: 2009, Annales Geophysicae 27, 3479.Lugaz, N., Vourlidas, A., Roussev, I.I., Morgan, H.: 2009, Solar Phys. 256, 269.

doi:10.1007/s11207-009-9339-4.Lugaz, N., Hernandez-Charpak, J.N., Roussev, I.I., Davis, C.J., Vourlidas, A., Davies, J.A.:

2010, Astrophys. J. 715, 493. doi:10.1088/0004-637X/715/1/493.Luhmann, J.G., Curtis, D.W., Schroeder, P., et al.: 2008, Space Sci. Rev. 136, 117.

doi:10.1007/s11214-007-9170-x.Maloney, S.A., Gallagher, P.T., McAteer, R.T.J.: 2009, Solar Phys. 256, 149.

doi:10.1007/s11207-009-9364-3.Mostl, C., Farrugia, C.J., Temmer, M., Miklenic, C., Veronig, A.M., Galvin, A.B., Leitner, M.,

Biernat, H.K.: 2009, Astrophys. Journ. Lett. 705, 180.Mostl, C., Temmer, M., Rollett, T., Farrugia, C.J., Liu, Y., Veronig, A.M., Leitner, M., Galvin,

A.B., Biernat, H.K.: 2010, Geophys. Res. Lett. 37, 24103. doi:10.1029/2010GL045175.Mostl, C., Rollett, T., Lugaz, N., et al.: 2011, Astrophys. J., 741, 34.

Richardson, I.G., Cane, H.V.: 2010, Solar Phys. 264, 189. doi:10.1007/s11207-010-9568-6.Robbrecht, E., Patsourakos, S., Vourlidas, A.: 2009, Astrophys. J. 701, 283. doi:10.1088/0004-637X/701/1/283.

Rollett, T., Mostl, C., Temmer, M., Veronig, A.M., Farrugia, C.J., Biernat, H.K.: 2012, SolarPhys. 276 293 doi:10.1007/s11207-011-9897-0.

Rouillard, A.P., Davies, J.A., Forsyth, R.J., et al.: 2008, Geophys. Res. Lett. 35. L10110.doi:10.1029/2008GL033767.

Rouillard, A.P., Savani, N.P., Davies, J.A., et al. 2009a, Solar Phys. 256, 307.doi:10.1007/s11207-009-9329-6.

Rouillard, A.P., Davies, J.A., Forsyth, R.J., et al.: 2009b, J. Geophys. Res. 114, 07106.doi:10.1029/2008JA014034.

Savani, N.P., Owens, M.J., Rouillard, A.P., Forsyth, R.J., Kusano, K., Shiota, D., Kataoka,R.: 2011, Astrophys. J. 731, 109. doi:10.1088/0004-637X/731/2/109.

Sheeley, N.R., Walters, J.H., Wang, Y.M., Howard, R.A.: 1999, J. Geophys. Res. 104, 24739.Tappin, S.J., Howard, T.A.: 2009, Space Sci. Rev. 147, 55. doi:10.1007/s11214-009-9550-5.Temmer, M., Rollett, T., Mostl, C., Veronig, A.M., Vrsnak, B., Odstrcil, D.: 2011, Astrophys.

J. 743, 101. doi:10.1088/0004-637X/743/2/101.Vourlidas, A., Howard, R.A.: 2006, Astrophys. J. 642, 1216. doi:10.1086/501122.Vrsnak, B., Zic, T., Falkenberg, T.V., Mostl, C., Vennerstrom, S., Vrbanec, D.: 2010, Astron.

Astrophys., 512 A43Williams, A.O., Davies, J.A., Milan, S.E., Rouillard, A.P., Davis, C.J., Perry, C.H., Harrison,

R.A.: 2009, Annales Geophysicae 27, 4359.Williams, A.O., Edberg, N.J.T., Milan, S.E., Lester, M., Franz, M., Davies, J.A.: 2011, J.

Geophys. Res. 116, 08103. doi:10.1029/2010JA015719.



21/22


Wood, B.E., Howard, R.A.: 2009, Astrophys. J. 702, 901. doi:10.1088/0004-637X/702/2/901.Wood, B.E., Howard, R.A., Plunkett, S.P., Socker, D.G.: 2009, Astrophys. J. 694, 707.

doi:10.1088/0004-637X/694/2/707.



22/22


heliospheric observations of stereo-directed coronal mass ejections in 2008--2010

Documents